nrk-subtitle-downloader.php

<?php

/**
 * Class NRK_Subtitle_Downloader
 *
 * This is a simple PHP script to help download subtitle files from NRK TV.
 * The subtitle files are in VTT-format which is supported by Plex.
 *
 * Usage:
 * php nrk-subtitle-downloader.php URL1,URL2,URL3 folder_path NOR
 *
 * URLs can be give like this:
 * https://tv.nrk.no/serie/broen/KMTE41004109/sesong-1/episode-1
 * KMTE41004109/sesong-1/episode-1
 */
class NRK_Subtitle_Downloader {

	/** @var array Array of URLs */
	private $urls = [];

	/** @var string Language string */
	private $language;

	/** @var string  */
	private $default_language = 'NOR';

	/** @var Path to folder for subtitle files */
	private $folder_path;

	/** @var class Colors class instance */
	private $colors;

	/** @var float Start time */
	private $startTime;

	/**
	 * NRK_Subtitle_Downloader constructor.
	 *
	 * @param bool $url_string
	 * @param null $folder_path
	 * @param null $language
	 */
	public function __construct( $url_string = false, $folder_path = null, $language = null ) {

	    // Check if folder path is specified
		$this->detectFolderPath( $folder_path );

		// Check if language is specified
	    $this->detectLangauge( $language );

		// Check if we should run on instantiation
		if ( $url_string ) {

			// Fetch subtitles
			$this->fetch( $url_string );

		}
	}

	/**
	 * Check if langauge is set.
	 *
	 * @param $language
	 */
	private function detectLangauge( $language ) {

		// Check if language is passed
		if ( $language ) {

			// Set language
			$this->setLanguage( $language );

		} else {

			// Set default language
			$this->setLanguage( $this->default_language );

		}

	}

	/**
	 * Check if folder path is passed.
	 *
	 * @param $folder_path
	 */
	private function detectFolderPath( $folder_path ) {

		// Check if folder path is passed
		if ( $folder_path ) {

			// Set folder path
			$this->setFolder( $folder_path );

		}

	}

	/**
     * Set language.
     *
	 * @param $language
	 */
	public function setLanguage( $language ) {
		$this->language = mb_strtoupper( $language );
	}

	/**
     * Set folder to store files in.
     *
	 * @param $path
	 */
	public function setFolder( $path ) {
		$this->folder_path = rtrim( $path, '/' ) . '/';
	}

	/**
     * Check if URL string contains multiple URLs.
     *
	 * @param $url_string
	 *
	 * @return bool
	 */
	private function isMultipleURL( $url_string ) {

		// Look for multiple URLs
		$http_count = substr_count( $url_string, 'http' );
		if ( $http_count > 1 ) {
			return true;
		}

		// Look for separator comma
		$comma_count = substr_count( $url_string, ',' );
		if ( $comma_count > 0 ) {
			return true;
		}

		return false;
	}

	/**
     * Parse URL string.
     *
	 * @param $url_string
	 */
	private function parseUrlString( $url_string ) {

		// Parse one or multiple URLs
		if ( $this->isMultipleURL( $url_string ) ) {
			$this->urls = explode( ',', $url_string );
		} else {
			$this->urls[] = $url_string;
		}

		// Remove empty ones
		$this->urls = array_filter( $this->urls, function( $value ) {
			return ! empty( trim( $value ) );
		} );

		// Remove any leading or trailing whitespace
		$this->urls = array_map( function( $value ) {
			$value = trim( $value );
			$value = rtrim( $value, '/' );
			return $value;
		}, $this->urls );

	}

	/**
	 * Ensure that destination folder exists.
	 */
	private function ensureFolder() {
		if ( $this->folder_path && ! file_exists( $this->folder_path ) ) {
			mkdir( $this->folder_path );
		}
	}

	/**
	 * Set start time.
	 */
	private function startTime() {
		$this->startTime = microtime(true);
    }

	/**
	 * Calculate execution time from start time.
	 *
	 * @return bool|string
	 */
    private function getExecutionTime() {
		if ( ! $this->startTime ) return false;
	    $endTime = microtime(true);
	    return number_format( $endTime - $this->startTime, 2 );
    }

	/**
	 * Initialized fetch process.
     *
	 * @param $url_string
	 */
	public function fetch( $url_string ) {

		// Set start time
	    $this->startTime();

		// Create new Colors class
		$this->colors = new Colors();

		// Ensure that destination folder exists
		$this->ensureFolder();

		// Parse URL string
		$this->parseUrlString( $url_string );

		// Fetch subtitle files by array
		$this->fetchByArray( $this->urls );

	}

	/**
     * Fetch subtitle files by array.
     *
	 * @param $array
	 */
	public function fetchByArray( $array ) {

		foreach ( $array as $raw_url ) {

			// Parse URL to object
			$clip = new NRK_Subtitle_Clip_Info( $raw_url, $this->language );
			
			if ( ! $clip->valid() ) {
				$this->fail( $raw_url );
				continue;
            }

			// Do remote fetch
			$content = $this->fetchSubtitleFile( $clip->url );
			if ( ! $content ) {

			    $this->fail( $clip->url, $clip->base_filename );
			    continue;

			}

			// Save subtitle file
			if ( ! $this->saveFile( $clip->filename, $content ) ) {

				$this->fail( $clip->url, $clip->base_filename );
				continue;

            }

			$this->success( $clip->url, $clip->base_filename );

		}

		$this->done();
	}

	/**
	 * Print completed message.
	 */
	private function done() {
		$executionTime = $this->getExecutionTime();
		echo $this->colors->getColoredString( sprintf( 'Process completed in %s seconds.', $executionTime ), 'green', null ) . "\n";
	}

	/**
     * Print success message.
     *
	 * @param $url
	 * @param $identifier
	 */
	private function success( $url, $identifier ) {
		echo $this->colors->getColoredString( sprintf( 'Successfully fetched subtitles from: %s (%s)', $url, $identifier ), 'green', null ) . "\n";
	}

	/**
     * Print failure message.
     *
	 * @param $url
	 * @param $identifier
	 */
	private function fail( $url, $identifier = false ) {
		$string = sprintf( 'Could not fetch subtitles from: %s', $url );
		if ( $identifier ) {
			$string .= ' (' . $identifier . ')';
		}
		echo $this->colors->getColoredString( $string, 'red', null ) . "\n";
	}

	/**
	 * Fetch remote file.
	 *
	 * @param $url
	 *
	 * @return bool|string
	 */
	private function fetchSubtitleFile( $url ) {
		$curl = curl_init();
		curl_setopt( $curl, CURLOPT_URL, $url );
		curl_setopt( $curl, CURLOPT_RETURNTRANSFER, true );
		curl_setopt( $curl, CURLOPT_HEADER, true );
		$response = curl_exec( $curl );
		$http_status_code = curl_getinfo( $curl, CURLINFO_HTTP_CODE );
		return ( ( ! empty( $response ) && $http_status_code === 200 ) ? $response : false );
	}

	/**
	 * Store subtitle file.
	 *
	 * @param $filename
	 * @param $content
	 *
	 * @return bool|int
	 */
	private function saveFile( $filename, $content ) {
		return file_put_contents( $this->folder_path . $filename, $content );
	}


}

/**
 * Class NRK_Subtitle_Clip_Info
 */
class NRK_Subtitle_Clip_Info {

	/** @var Assorted properties */
	private $full, $prefix, $number_1, $number_2, $number_3, $text_1, $season, $text_2, $episode, $id, $lang, $filename, $url, $base_filename;

	/** @var bool Valid indicator */
	private $valid;

	/** @var string RegEx pattern for video clip information */
	private $url_pattern = '/([A-Z]{4})([0-9]{2})([0-9]{2})([0-9]{4})\/([a-z-]+)([0-9]+)\/([a-z-]+)([0-9]+)$/';

	/** @var string Subtitle URL template for NRK */
	private $url_template = "https://undertekst.nrk.no/prod/%s%s/%s/%s/%s/%s.vtt";

	/** @var string Subtitle file template */
	private $subtitle_file_template = 'S%sE%s';

	/**
	 * NRK_Subtitle_Clip_Info constructor.
	 *
	 * @param $url
	 */
	public function __construct( $url, $lang ) {
		$this->lang = $lang;
		$this->valid = $this->readUrl( trim( $url ) );
		return $this->valid;
	}

	/**
	 * Check if valid.
	 */
	public function valid() {
		return $this->valid;
	}

	/**
     * Generate subtitle URL.
     *
	 * @return string
	 */
	private function generateSubtitleURL() {
		return sprintf(
			$this->url_template,
			$this->prefix,
			$this->number_1,
			$this->number_2,
			$this->id,
			$this->lang,
			$this->id
		);
	}

	/**
     * Populate class properties with data extracted from URL.
     *
	 * @param $matches
	 */
	private function populateProperties( $matches ) {

		// Extract variables
		list( $this->full, $this->prefix, $this->number_1, $this->number_2, $this->number_3, $this->text_1, $this->season, $this->text_2, $this->episode ) = $matches;

	}

	/**
     * Read unparsed URL.
     *
	 * @param $url
	 *
	 * @return bool
	 */
	private function readUrl( $url ) {

		// Parse URL into array
		preg_match( $this->url_pattern, $url, $matches );

		// Bail if extraction failed
		if ( count( $matches ) != 9 ) {
			return false;
		}

		// Populate properties
		$this->populateProperties( $matches );

		// Build full ID
		$this->id = $this->prefix . $this->number_1 . $this->number_2 . $this->number_3;

		// Generate subtitle URL
		$this->url = $this->generateSubtitleURL();

		// Generate subtitle filename
		$this->filename = $this->getSubtitleFilename();

		return true;

	}

	/**
     * Get file extension from remote URL.
     *
	 * @return mixed
	 */
	private function getFileExtension() {
		return pathinfo( $this->url, PATHINFO_EXTENSION );
    }

	/**
	 * Generate filename for subtitle file.
	 *
	 * @return string
	 */
	private function getSubtitleFilename() {

		// Add padding to season and episode number
		$season = str_pad( $this->season, 2, '0', STR_PAD_LEFT );
		$episode = str_pad( $this->episode, 2, '0', STR_PAD_LEFT );

		$this->base_filename = sprintf( $this->subtitle_file_template, $season, $episode );
		return $this->base_filename . '.' . $this->getFileExtension();
	}

	/**
	 * Get private property value.
	 *
	 * @param $name
	 *
	 * @return mixed
	 */
	public function __get( $name ) {
		return $this->{ $name };
	}

}

/**
 * Class Colors
 * Fecthed from: https://www.if-not-true-then-false.com/2010/php-class-for-coloring-php-command-line-cli-scripts-output-php-output-colorizing-using-bash-shell-colors/
 */
class Colors {
	private $foreground_colors = array();
	private $background_colors = array();

	public function __construct() {
		// Set up shell colors
		$this->foreground_colors['black'] = '0;30';
		$this->foreground_colors['dark_gray'] = '1;30';
		$this->foreground_colors['blue'] = '0;34';
		$this->foreground_colors['light_blue'] = '1;34';
		$this->foreground_colors['green'] = '0;32';
		$this->foreground_colors['light_green'] = '1;32';
		$this->foreground_colors['cyan'] = '0;36';
		$this->foreground_colors['light_cyan'] = '1;36';
		$this->foreground_colors['red'] = '0;31';
		$this->foreground_colors['light_red'] = '1;31';
		$this->foreground_colors['purple'] = '0;35';
		$this->foreground_colors['light_purple'] = '1;35';
		$this->foreground_colors['brown'] = '0;33';
		$this->foreground_colors['yellow'] = '1;33';
		$this->foreground_colors['light_gray'] = '0;37';
		$this->foreground_colors['white'] = '1;37';

		$this->background_colors['black'] = '40';
		$this->background_colors['red'] = '41';
		$this->background_colors['green'] = '42';
		$this->background_colors['yellow'] = '43';
		$this->background_colors['blue'] = '44';
		$this->background_colors['magenta'] = '45';
		$this->background_colors['cyan'] = '46';
		$this->background_colors['light_gray'] = '47';
	}

	// Returns colored string
	public function getColoredString($string, $foreground_color = null, $background_color = null) {
		$colored_string = "";

		// Check if given foreground color found
		if (isset($this->foreground_colors[$foreground_color])) {
			$colored_string .= "\033[" . $this->foreground_colors[$foreground_color] . "m";
		}
		// Check if given background color found
		if (isset($this->background_colors[$background_color])) {
			$colored_string .= "\033[" . $this->background_colors[$background_color] . "m";
		}

		// Add string and end coloring
		$colored_string .=  $string . "\033[0m";

		return $colored_string;
	}

	// Returns all foreground color names
	public function getForegroundColors() {
		return array_keys($this->foreground_colors);
	}

	// Returns all background color names
	public function getBackgroundColors() {
		return array_keys($this->background_colors);
	}
}

// CLI handling
$url         = ( isset( $argv[1] ) ? $argv[1] : false );
$destination = ( isset( $argv[2] ) ? $argv[2] : null );
$language    = ( isset( $argv[3] ) ? $argv[3] : null );
new NRK_Subtitle_Downloader( $url, $destination, $language );